Rebirth 08: Rise from copycat phones

Chapter 522: Half of the World’s Computing Power
Accompanied by the management of Zhiyun Guangcheng Branch, Xu Shenxue inspected in detail this largest science and technology park besides the headquarters.

The focus of the inspection was naturally the superconducting materials factory.

"We are currently mainly producing TK120 superconducting materials and secretly producing MK34 superconducting materials."

"What has been announced to the public is the production of TK120 superconducting materials, while the production of MK34 materials is kept confidential throughout the process. We have an independent factory, and all employees have been selected after strict screening. Everyone has signed a confidentiality agreement and voluntarily accepted all-round monitoring!"

"Our security department uses the security big data system to establish a security data file for every employee who can enter and access information related to MK34 materials, and conducts all-round monitoring to ensure that no problems will occur!"

Xu Shenxue listened to the explanation of the person in charge of the superconducting material factory and nodded slightly: "Safety and confidentiality are of paramount importance. We must attach great importance to them. You must cooperate well with the security department in this regard!"

The MK34 material was classified as top secret as soon as it was developed, and the relevant information was kept confidential in all aspects.

In fact, the protection of relevant information is not only the responsibility of Zhiyun Group’s internal security department. In fact, the relevant security agencies are more responsible. Zhiyun Group’s own security department is only responsible for internal confidentiality and preventing some fish from slipping through the net, and mainly cooperates with the security agencies.

Otherwise, the security department of Zhiyun Group alone cannot do this... Not to mention anything else, just the comprehensive monitoring of the many relevant personnel and their families involved in the MK34 materials is a hassle. In the past, there were relatively few people involved, but with the MK34 materials entering small-scale mass production and gradual application, more and more people know about it, and it has become increasingly difficult to keep it confidential.

In addition, while protecting information, many legal issues will also be involved!
These things are no longer something that can be handled by Zhiyun Group’s internal security department. They require a specialized security agency with high-level authority to do.

For now, the protection of relevant information of MK34 superconducting materials is still good, and there has been no information about MK34 materials circulating in the outside world before.

However, this large-scale information confidentiality operation will naturally attract the attention of hostile security agencies... Most people in the industry know that Guangcheng Science and Technology Park under Zhiyun Group has recently come up with a new thing, but they don’t know what this new thing is or what impact it will have.

The United States in particular was misled into thinking that Zhiyun Group had developed some carbon nanomaterials and was going to build a space elevator.

However, although MK34 material is already an excellent room-temperature superconducting material, it is still not enough... It can only superconduct at minus degrees Celsius and its expensive cost makes it impossible to use it in daily life. It can only be used in quantum computers, advanced medical care, scientific research and other fields, and it is far from being widely used!

Therefore, the R&D department of Zhiyun Group has been working hard to reduce the cost of MK34 materials, and trying to continue research and development based on the existing successful technical routes, trying to develop better room-temperature superconducting materials!
Superconducting materials at minus twenty-five degrees still have too many limitations...If you want to use them, you need a professional refrigeration equipment.

The research and development of new superconducting materials is still carried out by the Guangcheng Artificial Intelligence Materials Research Institute under Zhiyun Group.

His expertise is in developing the MK34 material, which enabled Yu Taihe to achieve rapid success.

In the laboratory inside the institute, Yu Taihe introduced the current research and development of superconducting materials to Xu Shenxue: "Mr. Xu, at present, most of the research and development efforts of our entire institute are focused on superconducting materials, including exploring large-scale low-cost mass production of MK34 materials, new applications of MK34 materials, and of course, more importantly, the research and development of new superconducting materials!"

"We have repeatedly conducted research and summarization based on the technical route of MK34 materials, and formed a technical route for carbon-based copper-iron composite superconducting materials. This is a new type of superconducting material technology route that has never appeared before!"

"We conducted in-depth research and study based on this technical route and preliminarily confirmed the basic characteristics of this technical route, which has very good room-temperature superconducting properties under normal pressure!"

"In fact, in addition to the MK34 material, we have also successfully developed three other carbon-based steel composite superconducting materials by changing the material formula and synthesis process. The superconducting temperature is between minus and minus degrees Celsius."

"And now, we are continuing our research and development, trying to develop a superconducting material with a higher temperature under normal pressure!"

"The MK76 project that I am currently working on with my team is just such a project. I have optimized the existing technical route. After completing the core theoretical framework of the new material last week, I started to hand it over to the computing center for big data calculations. Because the amount of calculations is too large and it needs to be continuously adjusted, we expect to complete the big data calculations next month and then enter the actual test stage!"

"In the first half of this year, our material model has undergone another important upgrade, and it is now getting easier and easier to use!"

As Yu Taihe spoke, he sighed that material research and development has become simple and easy now!

Although it still has a strong metaphysical flavor, it is at least much simpler and easier!

If his newly designed MK76 project follows the traditional materials research and development steps, such a complex composite material research would require hundreds of researchers and more than ten years of time just for the various complicated experiments in the early stages.

But now, this process can be handed over to the artificial intelligence scientific research assistance system. As long as there is enough computing power, it can be completed in about half a month.

This research and development process is an important feature of artificial intelligence materials science: it saves a lot of preliminary experimental time and costs.

Now Yu Taihe is engaged in research on artificial intelligence materials. He leads a group of subordinates to conduct theoretical innovation research on new materials. After designing theoretically possible technical directions, he hands them over to a specialized artificial intelligence scientific research system for a large professional material model to run.

When the large model is running, it will simulate thousands or even tens of thousands of different technical routes, and then find several technical routes that may be realized.

This saves a lot of time and costs.

In the research and development of many materials, the most time-consuming and labor-intensive research and development is actually the early stage of testing one by one... This research and development stage often requires hundreds or thousands of experiments before finally finding one of the successful ones...

This is the case when you are lucky. If you are unlucky, you may spend ten years and conduct tens of thousands of experiments without finding any usable materials.

In the past, developing a certain material might require hundreds of researchers to conduct various preliminary experiments for more than ten years. However, now that we use the material development big data model in this artificial intelligence scientific research system and hand over the data to the big model, we can basically get the results in ten days to half a month.

Of course... this actually comes at a price, which is that artificial intelligence will only try various technical routes according to the set parameters... it will not make mistakes, but it will not have a sudden inspiration either!
Mistakes and inspiration are very important links in the field of material research and development... Countless important materials were successfully developed because of mistakes or inspiration.

The role of this artificial intelligence scientific research system is not actually "groundbreaking research and development". In fact, it is used to reduce basic, repetitive, and mostly meaningless calculations, experiments and tests.

This kind of artificial intelligence research system can't replace top materials scientists like Yu Taihe, let alone the ordinary R&D workers under Yu Taihe.

The artificial intelligence scientific research system can only complete the initial low-tech repetitive calculations, experiments, and data statistics. The subsequent experiments and tests still need to be done by these highly educated R&D experts. Artificial intelligence cannot do this kind of thing.

Moreover, artificial intelligence has another big flaw, which is that it will not make mistakes, and without mistakes there will be no accidents, but materials science requires accidents...

No surprises, where does the new material come from?

Of course, even with many defects of one kind or another, the large material research and development model in the current artificial intelligence scientific research assistance system is still a very useful tool that can greatly reduce the cost and time of material research and development.

It can also allow material scientists to focus more on the innovation of the theoretical framework of new materials. After developing a new theory, they can design a theoretical model and directly throw it to big data to run. It is simple and convenient.

However, this also requires very high computing power!
This large material model within Zhiyun Group has been accumulated over many years and requires a huge amount of computing power to run!

The reason behind Yu Taihe's easy work in artificial intelligence materials science is the underlying artificial intelligence core algorithm developed by the Artificial Intelligence Research Institute. This algorithm combines the characteristics of materials science research and development to design a special material model.

The operating environment of this large material model also has very high requirements. Behind it is an extremely large computing center composed of tens of thousands of APO graphics cards and tens of thousands of GPUs.

This computing power is even greater than that used by many companies to train artificial intelligence.

Recently, the computing center for auxiliary material research and development has been upgraded. The Artificial Intelligence Research Institute has connected the computing center of this set of large material models to a new computing center, which has achieved a significant increase in computing power, which means faster computing.

Yu Taihe also knew something about this newly connected computing center. It was the world's first superconducting quantum computer array!

Zhiyun Group has been producing the YCL-34 superconducting quantum computer developed by MK51 superconducting materials in small batches. Some of them are supplied to related confidential research and development institutions, but more of them are used by the group itself to build superconducting quantum computer arrays. The first superconducting quantum computer array was successfully built at the beginning of this year. This supercomputing center, which consists of more than 51 YCL-6000 superconducting quantum computers, more than 320 AI graphics cards, and tens of thousands of WZ server GPUs, has become the most powerful computing center under Zhiyun Group once it was built.

By the way, current superconducting quantum computers cannot complete all types of computing tasks for the time being because the number of controllable quantum bits is still not high, so traditional computers are still needed to cooperate in calculations.

Despite some defects, superconducting quantum computers still play a huge role in scientific research. When performing computing tasks in some scientific research fields, their efficiency is hundreds or even thousands of times that of traditional computers.

This new computing center, which is composed of YCL51 superconducting quantum computers, AI6000 graphics cards, and WZ320 server CPUs, has a computing power that accounts for % of the total computing power of all iCloud Group.

This is quite powerful data.

You should know that Zhiyun Group has been continuously building computing centers for the field of artificial intelligence many years ago. Its server GPU graphics cards have been updated from AI1000 to the current AI6000...

After using the latest AI6000, the old graphics cards such as AI5500, AI5000, AI4500 and AI4000 are still in use.

The numerous large-scale computing centers accumulated over the years have enabled Zhiyun Group to possess the world's largest computing resources. The computing resources of ten Googles combined are not as much as the computing power of Zhiyun Group alone.

In fact, most of the server GPU chips produced by Zhiyun Group are used by the company itself...less than 50% are sold.

However, the computing power of this new computing center is equivalent to 40% of Zhiyun Group’s existing computing power, which shows how powerful it is.

Zhiyun Group is also building a second superconducting quantum computer comprehensive computing center. This second computing center is deployed in the southwest region and has even greater computing power. It is expected to use more than 51 YCL- superconducting quantum computers.

Even the AI600 graphics cards used as a combination exceed 320 pieces, in addition to a huge number of WZ server CPUs.

The second superconducting quantum computer computing center is expected to be put into use early next year, by which time Zhiyun Group's computing power will be further greatly enhanced.

It’s not just that we are far ahead of other companies, but we are far ahead of the total computing power of other companies in the world!

In fact, we are now far ahead. If there are 100 computing power resources used in the field of artificial intelligence worldwide, Zhiyun Group alone accounts for at least 50%.

Of the remaining 50 percent, other domestic companies occupy a large share, and related research and development institutions occupy a considerable share. For example, a large number of computing power centers are needed in the field of national defense, and some domestic Internet companies are also building computing power centers on a large scale to deploy GTAI or self-developed artificial intelligence.

Thanks to the geographical advantages, relevant domestic institutions or enterprises can easily purchase Zhiyun Group's computing power cards nearby, so it is very convenient.

If this part is also taken into account, then domestic computing power resources can probably account for about 70% of the world's total computing power.

More than 20% of the computing power resources are bought by American companies... They spend hundreds of billions of dollars every year to purchase Zhiyun Group's sky-high priced computing power cards. This money is not wasted, but they really buy huge computing power.

As for why we haven’t developed top-level artificial intelligence yet, especially artificial intelligence for terminal fields, this is a problem with the underlying algorithms of artificial intelligence... We have the computing power, but it’s useless if we can’t figure out the underlying algorithms.

I have a bad brain, who can I blame...

As for companies in other countries, their computing power is very small. Apart from China and the United States, the computing power of all other countries combined is estimated to be only seven or eight points, and they are also scattered in Japan, South Korea and Europe.

India, which claims to be a major power in artificial intelligence, is not ranked high in the world in terms of computing resources.

Look at the Americans. Their many companies combined have to purchase millions of APO graphics cards from Zhiyun Group every year.

For APO5000 graphics cards alone, the total order volume from American companies has now reached millions of pieces, which has even put the production line of APO5000 graphics cards of Zhiyun Microelectronics, a subsidiary of Zhiyun Group, into a state of panic.

Look at the Americans' courage to spend money. Even the senior management of Zhiyun Group, a competitor, have to say: admiration!

These American companies are really bold in the field of artificial intelligence and are not stingy when it comes to spending money.

Hundreds of billions or even hundreds of billions of dollars are being poured down without even blinking an eye!
Let’s take a look at the various quibbles that Indian customers are making when they purchase tens of thousands of APO5000 graphics cards… Now the sales staff of Zhiyun Group’s computing power card business department are annoyed when they see Indian customers: they like to make a fuss even though the purchase volume is so small, and they want to bargain for a refund one moment, and the next they want to go through all kinds of trouble to get a refund.

One moment they say that the Indian market has huge potential and is the center of future human artificial intelligence. If your Zhiyun Group graphics cards want to enter the country, you must set up a factory there, or else you will be taxed 100% on your Zhiyun graphics cards.

Humph, just ask you if you are afraid!
This made the head of Zhiyun Group's GPU business department speechless. Where did you get the courage? You were the one begging me to sell you the graphics card, not me begging you to sell it... Let's get this clear!

They also imposed tariffs on my graphics cards. A large number of CEOs of high-tech companies in the United States dare not say this in front of me...

As a result, the head of the GPU business unit in Zhiyun Group, without much hesitation, postponed the order for APO5000 graphics cards from Indian companies to the year after next.

These tens of thousands of pieces of reserved production capacity were freed up to fill new orders from several Japanese companies.

The head of the Japanese company accepted this generous gift with a smile!

This new order, which has just been placed, will probably be delivered by the year after next according to normal scheduling. After all, APO5000 graphics cards are in short supply, and it is normal to wait in line for a year or even two years to get the goods.

However, they did not expect that Zhiyun’s people suddenly informed them that a batch of production capacity scheduled for next autumn could be freed up, and asked them to prepare funds in advance so that they could take delivery of the goods next autumn.

What a good thing!
After careful investigation, we found out that it was the Indians who were causing trouble and pissed off Zhiyun, so the boss of Zhiyun's GPU business unit personally adjusted the production capacity and gave them the production capacity originally intended for Indian companies.

This Indian is a good person and does good things!
India has made a lot of jokes in the field of artificial intelligence, but they are really working on artificial intelligence... After all, they are planning to develop fifth-generation and sixth-generation aircraft, so developing artificial intelligence is not a surprising thing.

At worst, there is GTAI as a backup... Even if you use the open source GTAI and slightly change the interface and fonts, and call it India's own artificial intelligence, few Indians can tell the difference.
From the global distribution of computing resources, we can see that these days, the companies engaged in artificial intelligence are basically from China and the United States, and companies from other countries have no chance.

This doesn't mean that Zhiyun Group doesn't sell APO graphics cards. Now Zhiyun Group basically has an open supply attitude towards APO graphics cards and will sell them to you as long as you pay!
The problem is that many companies in other countries don’t have much money and can’t afford to sell large quantities of computing power cards.

The most important thing is that after they bought the APO graphics card, it was of no use...even using open source AI like GTAI does not require so much computing power.

Because they don’t even have many decent Internet companies…

------

The underlying core algorithms, computing centers and massive amounts of data are the three core factors in artificial intelligence.

In these three fields, Zhiyun Group ranks among the top in the world, far ahead of other companies.

Especially the most core algorithm field.

However, people cannot see the algorithm and it is difficult to evaluate it.

What is better evaluated is the computing power!
Computing power is a hard indicator and it is clear at a glance!
And Zhiyun Group alone occupies more than half of the world's computing power resources!
(End of this chapter)